1 Re-engineering Engineering Vinod Khosla Kleiner Perkins Caufield & Byers [email protected] Sept 2000

1

Re-engineering Engineering Re-engineering Engineering

Vinod KhoslaVinod KhoslaKleiner Perkins Caufield & ByersKleiner Perkins Caufield & Byers

[email protected]@kpcb.com

Sept 2000Sept 2000

2

“…“…every strategic inflection point [is] every strategic inflection point [is] characterized by a ‘10X’ change …”characterized by a ‘10X’ change …”

““There’s wind and then there is a typhoon,There’s wind and then there is a typhoon,there are waves and then there’s a tsunami” there are waves and then there’s a tsunami”

- Andy Grove- Andy Grove

There’s Change And ThenThere Is Change!

3

he Environmenthe Environment

4

The EnvironmentThe Environment

Hyper efficiency or Adam Smith IIHyper efficiency or Adam Smith II Hyper speedHyper speed Winner take all economyWinner take all economy Value add transparencyValue add transparency Risk as a requirement?Risk as a requirement? Change as a processChange as a process Technology as a driverTechnology as a driver Diseconomy of ScaleDiseconomy of Scale Technostructure & InfostructureTechnostructure & Infostructure

5

Environment: “Change as a Process”Environment: “Change as a Process”

Business model evolution everyday!Business model evolution everyday! Infrastructure renovationInfrastructure renovation Systems evolutionSystems evolution Strategy evolutionStrategy evolution

6

Environment : Technology as a Strategy DriversEnvironment : Technology as a Strategy Drivers

7

Environment : Diseconomies of ScaleEnvironment : Diseconomies of Scale

Diseconomies of process / hierarchyDiseconomies of process / hierarchy

Time for information disbursementTime for information disbursement

Lack of “real time” – poor cost & poor Lack of “real time” – poor cost & poor

partner experiencepartner experience

The six month “learning curve” edge in a The six month “learning curve” edge in a

short product cycle world!short product cycle world!

8

Technostructure & InfostructureTechnostructure & Infostructure

Specialization & complexity of technologySpecialization & complexity of technology Decision-making: top down or bottom up?Decision-making: top down or bottom up? The role of the “fringe” employeeThe role of the “fringe” employee Nuances as pitfalls Nuances as pitfalls Horizontal & vertical communicationHorizontal & vertical communication

& cooperation - & cooperation - notnot top down top down Information based, dynamic decision Information based, dynamic decision

makingmaking

9

The Environment -It’s Not About a “Website”The Environment -It’s Not About a “Website”

Brand building – pull vs. push vs. informationBrand building – pull vs. push vs. information Product marketing & research- a new paradigm?Product marketing & research- a new paradigm? Cost of services and goodsCost of services and goods New variables– customer database, personalized New variables– customer database, personalized

productproduct Logistics (Fed Ex), buying (FreeMarkets), admin.Logistics (Fed Ex), buying (FreeMarkets), admin. Operations: ordering, billing, service,Operations: ordering, billing, service,

returns, inventory...returns, inventory... TransparencyTransparency New models- eLance, eBay, Priceline, AmazonNew models- eLance, eBay, Priceline, Amazon

10

New Networks, New PossibilitiesNew Networks, New Possibilities

Internet data centers Internet data centers

High bandwidth last mileHigh bandwidth last mile

Total bandwidth explodingTotal bandwidth exploding

Connectivity - “evernet”, “everyplace”Connectivity - “evernet”, “everyplace”

Applications over IP-VPNsApplications over IP-VPNs

ASP’sASP’s

11

he CIO’s Issueshe CIO’s Issues

12

CIO’s IssuesCIO’s Issues

The problem of legacy - systems, people,...The problem of legacy - systems, people,... Skills shortageSkills shortage Re-engineering the enterprise for Re-engineering the enterprise for technology technology

basedbased competition/strategy competition/strategy Intranets & extranets among islands of Intranets & extranets among islands of

information/systemsinformation/systems Dynamic information architectureDynamic information architecture vs. vs.

static databases (“enterprise models”)static databases (“enterprise models”) Real timeReal time corporation & future of software corporation & future of software New application proliferationNew application proliferation

13

CIO’s Issues: Legacy EngineeringCIO’s Issues: Legacy Engineering

Optimization for whatOptimization for what

– CostCost

– PerformancePerformance

– ReliabilityReliability

SystemsSystems

Business ProcessBusiness Process

14

CIO’s Issues: Skills ShortageCIO’s Issues: Skills Shortage

Complexity increasing exponentiallyComplexity increasing exponentially– More systemsMore systems– More applicationsMore applications– More devicesMore devices

Rapid changeRapid change– Faster versionsFaster versions– New requirementsNew requirements

Human capitalHuman capital– Linear growth of supplyLinear growth of supply– Outflow from MISOutflow from MIS

15

CIO’s Issues: Change ManagementCIO’s Issues: Change Management

Old databasesOld databases

Old systemsOld systems

Legacy logicLegacy logic

C/S architecturesC/S architectures

New applicationsNew applications

New usersNew users

New “internet” New “internet” environmentenvironment

Multi-architecture Multi-architecture systemssystems

16

CIO’s Issues: Real-Time CorporationCIO’s Issues: Real-Time Corporation

Manual processesManual processes

Batch processesBatch processes

Disconnected systemsDisconnected systems

Limited “partner” connectivity (EDI)Limited “partner” connectivity (EDI)

Increasing “real time” requirementsIncreasing “real time” requirements

17

CIO’s Issues: Information ArchitectureCIO’s Issues: Information Architecture

IntranetsIntranets ExtranetsExtranets Multi-architecture environmentMulti-architecture environment PersonalizationPersonalization ManageabilityManageability Business strategy ready infrastructureBusiness strategy ready infrastructure

• Unified customer data (Epiphany)Unified customer data (Epiphany)• Partner information integrationPartner information integration

(contract manufacturer)(contract manufacturer)• Billing & moreBilling & more

18

CIO’s Issues: Engineering MethodologyCIO’s Issues: Engineering Methodology

EvolvabilityEvolvability SpecializationSpecialization ExperimentationExperimentation Change isolationChange isolation DiversityDiversity Connectivity orientedConnectivity oriented Best of breed orientedBest of breed oriented StandardsStandards

19

he Road Aheadhe Road Ahead

20

The Road Ahead ...The Road Ahead ...


New goalsNew goals

ReliabilityReliability

Gene poolGene pool

Process - organized chaos?Process - organized chaos?

21


Internet scale data centers Internet scale data centers

High bandwidth (really!) last mileHigh bandwidth (really!) last mile Total bandwidth explosion Total bandwidth explosion (Continues…)(Continues…)

Connectivity - “evernet”, “everyplace”, Connectivity - “evernet”, “everyplace”, “everydevice” (wireless plus more)“everydevice” (wireless plus more)

Applications over IP, wireless, …Applications over IP, wireless, …transitiontransition

ASP’sASP’s

22

Road Ahead: “New” GoalsRoad Ahead: “New” Goals

Complexity thru Complexity thru federationfederation NOT integration NOT integration

Adaptability & evolvabilityAdaptability & evolvability

ConfigurableConfigurable NOT customized NOT customized

Modularity – “Modularity – “micro” openmicro” open systems model systems model

PersonalizationPersonalization

Application interoperability, unified UIApplication interoperability, unified UI

Dramatically new management systemsDramatically new management systems

23

Road Ahead : A “new” ReliabilityRoad Ahead : A “new” Reliability

The shuttle Challenger: designed not to The shuttle Challenger: designed not to failfail

Biological systems: designed to fail Biological systems: designed to fail gracefullygracefully

Complex systems: “evolutionary Complex systems: “evolutionary approach”approach”

24/7 mission critical systems (Routers vs. 24/7 mission critical systems (Routers vs. phone network)phone network)

24

Road Ahead: Diverse “Gene Pool”Road Ahead: Diverse “Gene Pool”

Mix of skillsMix of skills

Mix of areasMix of areas

Mix of personalitiesMix of personalities

Mix of biasesMix of biases

25

The Shepherd or the Sargent?The Shepherd or the Sargent?

Road Ahead: Organized Chaos “process”Road Ahead: Organized Chaos “process”

The flakes vs. architects vs. implementorsThe flakes vs. architects vs. implementors

ExperimentationExperimentation

ExecutionExecution

Budgets, schedules, tasks vs. project stageBudgets, schedules, tasks vs. project stage

26

ew Areas for Innovationew Areas for Innovation

27

New Areas for InnovationNew Areas for Innovation

““Extranet information architecture”Extranet information architecture”

Virtual computerVirtual computer

Network services - Network services - “decomposing” the computer“decomposing” the computer

28

New Area: “Ibase” for the “Extranet”New Area: “Ibase” for the “Extranet”

Multi-architecture architecturesMulti-architecture architectures Messaging paradigmsMessaging paradigms Heterogeneous databasesHeterogeneous databases MetadataMetadata Entitlement: authentication, authorization...Entitlement: authentication, authorization... Inter-enterprise MIS, diverse environmentsInter-enterprise MIS, diverse environments TranslationTranslation EAIEAI Connectivity- speeds, modes, devicesConnectivity- speeds, modes, devices

The Extranet Information ArchitectureThe Extranet Information Architecture

29

Case Study: Asera “IBASE”Case Study: Asera “IBASE”

Architecture for multi-architecture Architecture for multi-architecture integrationintegration

Unification of the UI - “personal portal”Unification of the UI - “personal portal” Universal, application independent Universal, application independent

“entitlement system”“entitlement system” Messaging, EAI, translation…Messaging, EAI, translation… Customizable and personalizedCustomizable and personalized Not quite flexible “business object modeling”Not quite flexible “business object modeling” Not quite the work flow and rules engineNot quite the work flow and rules engine

30

Architecture connecting architectures...Architecture connecting architectures...

31

32

Single sign-on

Back OfficeERP

SFA

CRM

Catalog

NewsMessageBoards

SecurityEAIEAI

Profile

EAI

Profile

User workflow

33

New Area: “Virtual Computer” New Area: “Virtual Computer”

Scalability of hardware - add & deleteScalability of hardware - add & delete

Self managementSelf management

Geographic distributionGeographic distribution

Load balancing, caching, COS, … servicesLoad balancing, caching, COS, … services

““Network operating system” for the IBASENetwork operating system” for the IBASE

A Computer Distributed Over the InternetA Computer Distributed Over the Internet

34

Case Study: Router NetworksCase Study: Router Networks

Behave as “one” machineBehave as “one” machine

System self-adjusts to “node” failuresSystem self-adjusts to “node” failures

Capacity can be added/deleted - “self Capacity can be added/deleted - “self

organizing”organizing”

Geographically disbursedGeographically disbursed

35

New Areas: Network ServicesNew Areas: Network Services

Storage servicesStorage services Database servicesDatabase services Web servers/HTTP serversWeb servers/HTTP servers TCP/IP session serversTCP/IP session servers Application serversApplication servers Composite servicesComposite services

– ReplicationReplication– Load balancingLoad balancing– DistributionDistribution

The “decomposed” Computer ArchitectureThe “decomposed” Computer Architecture

36

Case Study: ZambeelCase Study: Zambeel

……distributed data distributed data servicesservices

37

Databases Changing NeedsDatabases Changing Needs

Re-engineering DatabasesRe-engineering Databases

Data vs. “other”Data vs. “other” MetadataMetadata QOS, reliability vs cost vs. access timeQOS, reliability vs cost vs. access time DistributionDistribution Scalability: size, concurrent users ...Scalability: size, concurrent users ... SecuritySecurity

38


Entitlement system, securityEntitlement system, security Changing hardware – routers as a model of Changing hardware – routers as a model of

HW/SW interaction & evolutionHW/SW interaction & evolution Changing application environmentsChanging application environments ““Need” diversity: traditional databases, content, Need” diversity: traditional databases, content,

portal data, evolution, “free web” vs. corporateportal data, evolution, “free web” vs. corporate Database use: tables, relations, emergent Database use: tables, relations, emergent

behavior, analysis types, browsing, version behavior, analysis types, browsing, version control, logic vs. datacontrol, logic vs. data

Databases Changing NeedsDatabases Changing Needs

39


Large enterprises have hundreds, sometimes thousands, ofLarge enterprises have hundreds, sometimes thousands, oflarge-scale, complex packaged and custom applications. large-scale, complex packaged and custom applications. Interoperation between these applications is essential for the Interoperation between these applications is essential for the flexibility needed by enterprises to introduce new web-based flexibility needed by enterprises to introduce new web-based applications services, meet regulator requirements, reduce time applications services, meet regulator requirements, reduce time to market, reduce costs, and execute business mergers.to market, reduce costs, and execute business mergers.

Because of gizmos, we foresee an explosion in the size and scale Because of gizmos, we foresee an explosion in the size and scale of data clients and servers -- trillions of gizmos will need billions of data clients and servers -- trillions of gizmos will need billions of servers. The number, mobility, and intermittent connectivity of servers. The number, mobility, and intermittent connectivity of gizmos render current client-server and three-tier software of gizmos render current client-server and three-tier software architectures unsuitable for supporting such devices. Most architectures unsuitable for supporting such devices. Most gizmos will not have a user interface and cannot have a database gizmos will not have a user interface and cannot have a database administrator -- they must be self-managing, very secure, and administrator -- they must be self-managing, very secure, and very reliable.very reliable.

Database Needs Database Needs Asilomar Report on Database (12/98)Asilomar Report on Database (12/98)

40

““Plug & Play Database Management Systems ReportPlug & Play Database Management Systems ReportAsilomar Report on Database (12/98)Asilomar Report on Database (12/98)


No knobs operation; a no-knobs system must adapt as conditions changeNo knobs operation; a no-knobs system must adapt as conditions change Have the database system automatically discover and interact with the other Have the database system automatically discover and interact with the other

database systems accessible on the network. This information discovery process database systems accessible on the network. This information discovery process will require that database systems provide substantially more metadata that will require that database systems provide substantially more metadata that describes the meaning of the objects they manage.describes the meaning of the objects they manage.

Billions of web clients will be accessing millions of databases. Enterprises will set Billions of web clients will be accessing millions of databases. Enterprises will set up large-scale federated database systems.up large-scale federated database systems.

Imprecise information will not only appear as the output of queries; it already Imprecise information will not only appear as the output of queries; it already appears in data sources as wellappears in data sources as well

Most organizations need continuous system operation. Designing a software Most organizations need continuous system operation. Designing a software system that never fails requires remote replicas and dynamic reconfiguration. system that never fails requires remote replicas and dynamic reconfiguration. Made to not fail or fail gracefully?Made to not fail or fail gracefully?

Changing needs at the app level and changing hardware environment –Changing needs at the app level and changing hardware environment –no assumptionsno assumptions

The information utility: make it easy for everyone to store, organize, access,The information utility: make it easy for everyone to store, organize, access,and analyze the majority of human information onlineand analyze the majority of human information online

41

Case Studies - “Issues”Case Studies - “Issues”

OracleOracle

– ““Poor” implementationsPoor” implementations

– Is Oracle or the implementation the problem?Is Oracle or the implementation the problem?

– People or technology the invariant problem?People or technology the invariant problem?

Windows/NTWindows/NT

– Slow innovationSlow innovation

– One size fits allOne size fits all

– Not very flexibleNot very flexible

42

Case Study: “Solutions”- LinuxCase Study: “Solutions”- Linux

ModularityModularity

High customizabilityHigh customizability

Higher reliabilityHigher reliability

Short “version cycle”Short “version cycle”

43

LinuxLinux

……engineering from the bottom upengineering from the bottom up

……powerful testimony to the power of evolutionpowerful testimony to the power of evolution

……the invisible hand of Adam smiththe invisible hand of Adam smith

44

Linux: Cathedral and the BazaarLinux: Cathedral and the Bazaar

Ideas are many…..the trick is to recognize the good ones.Ideas are many…..the trick is to recognize the good ones.

Be ModularBe Modular

Be flexible: Change your mind and approach. Be flexible: Change your mind and approach.

Release early and oftenRelease early and often

Peer review is essentialPeer review is essential

Feedback is key to rapid and effective code development and Feedback is key to rapid and effective code development and

debugging. debugging.

Beta testers are your most valuable resourceBeta testers are your most valuable resource

Eric Raymond:

45

Evolvable Systems (Sharky)Evolvable Systems (Sharky)

Centrally designed protocols start out strong and Centrally designed protocols start out strong and improve logarithmically….evolvable protocols start improve logarithmically….evolvable protocols start up weak and improve exponentiallyup weak and improve exponentially

•Only solutions that produce partial results

when partially implemented are evolvable

•What is, is wrong

•Evolution is cleverer than you are

46

Linux has dazzled engineers, users, andLinux has dazzled engineers, users, andcritics alike with its immense complexitycritics alike with its immense complexityand dazzling performance. Its existenceand dazzling performance. Its existenceowes as much to accidental luck as toowes as much to accidental luck as toingenious hack. It is a story of somethingingenious hack. It is a story of somethingout of nothing, a powerful testimony toout of nothing, a powerful testimony tothe power of evolution.the power of evolution.

Analytically, Linux is twiceAnalytically, Linux is twiceimprobably - once for itsimprobably - once for itstechnical complexity, andtechnical complexity, andtwice for its social complexity.twice for its social complexity.

Given the essentialGiven the essentialingredients of evolution …ingredients of evolution …any system, natural orany system, natural orartificial, can evolve into aartificial, can evolve into acomplex design throughcomplex design throughincremental changesincremental changesexplored in parallel.explored in parallel.

Linux: A Bazaar at the Edge of ChaosLinux: A Bazaar at the Edge of Chaos

47

It is not individualIt is not individualefficacy that propelsefficacy that propelscollective action, butcollective action, butgroup efficacy thatgroup efficacy thatmotivates individualmotivates individualcontribution.contribution.

The Linux project has neitherThe Linux project has neithertop-down planning nor atop-down planning nor acentral body vested withcentral body vested withbinding and enforcingbinding and enforcingauthorities. Its power, theauthorities. Its power, thesource of its bubblingsource of its bubblingcreativity, is instead in thecreativity, is instead in theceaseless interactivity amongceaseless interactivity amongits developers.its developers.

Coordination is a crucialCoordination is a crucialelement sustaining collectiveelement sustaining collectiveefforts, giving the Linuxefforts, giving the Linuxproject its integrity thatproject its integrity thatunfolds the seemingly chaoticunfolds the seemingly chaoticyet infinitely creativeyet infinitely creativeprocess of evolution.process of evolution.


48

Given the essentialGiven the essentialingredients of evolution …ingredients of evolution …any system, natural orany system, natural orartificial, can evolve into aartificial, can evolve into acomplex design throughcomplex design throughincremental changesincremental changesexplored in parallel.explored in parallel.

Providing a public good isProviding a public good isnot simply a matter ofnot simply a matter ofpooling efforts, but also ofpooling efforts, but also ofcoordinating efforts.coordinating efforts.


49

Linux: Personal ViewsLinux: Personal Views

Powerful Development MechanismsPowerful Development Mechanisms

Methodology : Ultimate “open system”Methodology : Ultimate “open system”

Maybe evolutionary “biological” systems Maybe evolutionary “biological” systems

approach works?approach works?

50

Personal Views : Development MechanismsPersonal Views : Development Mechanisms

Modular developmentModular development

Successive refinementSuccessive refinement

Aggressive peer reviewAggressive peer review

Forced “Architecture, Architecture, Forced “Architecture, Architecture,

Architecture”Architecture”

51

Personal Views : Methodology Ultimate “Open System”

Personal Views : Methodology Ultimate “Open System”

Origin of “open systems” circa 1982Origin of “open systems” circa 1982

Methodologically forced opennessMethodologically forced openness

Methodologically forced modularityMethodologically forced modularity

Methodologically forced adaptabilityMethodologically forced adaptability

52

Personal Views : Biological System?Personal Views : Biological System?

““Emergent”?Emergent”?

Biological resiliency?Biological resiliency?

Biological, incremental evolution?Biological, incremental evolution?

Defects cause variation/experimentation?Defects cause variation/experimentation?

Complexity theory proof?Complexity theory proof?

53

Reading:Reading:

The Cathedral and the Bazaar (Eric Raymond)The Cathedral and the Bazaar (Eric Raymond)

In Praise of Evolvable Systems (Clay Shirky)In Praise of Evolvable Systems (Clay Shirky)

The Circus Midget and the Fossilized Dinosaur The Circus Midget and the Fossilized Dinosaur

Turd (Martin Hock)Turd (Martin Hock)

Linux: A Bazaar at the Edge of Chaos (Ko Linux: A Bazaar at the Edge of Chaos (Ko

Kusabara) Kusabara)

54

EconomicsEconomics

The relative cost of computing and human attention has The relative cost of computing and human attention has changedchanged

This new economics requires that computer systems be This new economics requires that computer systems be autoeverything: autoinstalling automanaging, autoeverything: autoinstalling automanaging, autohealing, and autoprogramming autohealing, and autoprogramming

Computers can augment human intelligence by Computers can augment human intelligence by analyzing and summarizing data,analyzing and summarizing data,by organizing it, by intelligently answering direct by organizing it, by intelligently answering direct questions and by informing people when interesting questions and by informing people when interesting things happenthings happen

The cost of “failure” is increasing exponentiallyThe cost of “failure” is increasing exponentially

55

The Weather Forecast …The Weather Forecast …

Rate of change will accelerate - life will be Rate of change will accelerate - life will be more complex, busier . . . more complex, busier . . .

Adaptability, agility & momentum will be Adaptability, agility & momentum will be the key to success!the key to success!

Innovation, opportunities & Innovation, opportunities & entrepreneurship will thriveentrepreneurship will thrive

Disruption will be the order of the dayDisruption will be the order of the day Fun, fortunes & failure will be in Fun, fortunes & failure will be in

abundanceabundance

Comments?Comments?Resumes?Resumes?

Business Plans?Business Plans?

[email protected]/team/vinod.htmlwww.kpcb.com/team/vinod.html

mailto:[email protected]

57

Impact on Software ArchitectureImpact on Software Architecture

New tradeoffs for the new, diverse New tradeoffs for the new, diverse networked environmentnetworked environment

Architecture for integration -- plug and Architecture for integration -- plug and play application architectureplay application architecture

Configurable and personalizedConfigurable and personalized Rapid deploymentRapid deployment Ease of change managementEase of change management Continuous availabilityContinuous availability

58

Application Challenges Application Challenges

Flexible business object modelingFlexible business object modeling Business workflow enginesBusiness workflow engines

– Rapid business process modelingRapid business process modeling– Rapid business process changeRapid business process change– Intranet/Extranet enabledIntranet/Extranet enabled

Business rule enginesBusiness rule engines– Complex business processing rulesComplex business processing rules– Customization/profiling rulesCustomization/profiling rules

Doing all these right!Doing all these right!

59

1999 Annual Report on KPCB 3501999 Annual Report on KPCB 350 99 Stats99 Stats

Total RevenueTotal Revenue$80+ billion$80+ billionEmploymentEmployment 195,000195,000Market ValueMarket Value $658 billion$658 billion

1999 Annual Report on KPCB 3501999 Annual Report on KPCB 350 99 Stats99 Stats

Total RevenueTotal Revenue$80+ billion$80+ billionEmploymentEmployment 195,000195,000Market ValueMarket Value $658 billion$658 billion

KPCBKPCB Founding investors in 11 of Forbes 500: Founding investors in 11 of Forbes 500:

– Sun, Compaq, Genentech, AOL, LSI Logic,Sun, Compaq, Genentech, AOL, LSI Logic,Ascend, Juniper, Cerent, Netscape, Amazon,Ascend, Juniper, Cerent, Netscape, Amazon,Excite, @HOME ...Excite, @HOME ...

Not financiers: technologists, operating execsNot financiers: technologists, operating execs

60Share Price on 1/31/00Share Price on 1/31/00

KP Net Services LeadershipKP Net Services Leadership

$230B42M Users#1 Revenue

$230B42M Users#1 Revenue

$22B20M Users

#1 E-commerce

$22B20M Users

#1 E-commerce

$14B0.2M Subs 19M Users

#1 Broadband #6 Audience

$14B0.2M Subs 19M Users

#1 Broadband #6 Audience

$11B12M Users

#1 Revenue SW

$11B12M Users

#1 Revenue SW

$6.8B#1 Online Home

Related Site

$6.8B#1 Online Home

Related Site$9.5B#1 Healthcare

Site

$9.5B#1 Healthcare

Site

$1.3B#1 OnlineDrugstore

$1.3B#1 OnlineDrugstore

#1 Online Grocer#1 Online Grocer

$4.2B#1 CRMSoftware

$4.2B#1 CRMSoftware

$1.7B#1 Online

Credit Card

$1.7B#1 Online

Credit Card

Total ShareholderValue: $300B +

61

Kleiner Perkins B2B LeadershipKleiner Perkins B2B Leadership

IronPlanet

Vertically focused

62

KP’s Long History of Building Big, KP’s Long History of Building Big, Relevant CompaniesRelevant Companies

63

64

65

……

66

Documents

1 Re-engineering Engineering Vinod Khosla Kleiner Perkins Caufield & Byers [email protected] Sept 2000